Ant
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ANT-100-1 (complete)
1.12NAUC
4
ANT-50-1 complete
91NAUC
4
ANT 25-1 (complete)
77NAUC
4
Ant discrete-regime, exponential dwell
20,679Mean Evaluation Return
4
Ant Extreme Gravity-Shift
70.8Ant-m (m)
4
Ant
94.6TDSR
4
Ant v5
5,608.2Mean Episode Return
4
Ant
15.8Value Estimation MAE
4
Ant v5
890Episodes to Threshold 2500
4
Ant Default
3,672.6Average Return
4
ant 10-p
0.927Normalized Return
4
ant 2-p
146.1Normalized Return
4
ant 4-p
136Normalized Return
4
Ant target domain 1.5 density
6,036Reward
4
Ant 0.5 gravity (target)
3,380Total Reward
4
Ant 1.5 gravity MuJoCo (test)
5,320Reward
4
Ant MuJoCo (broken source)
6,183Reward
4
Disabled-Ant (meta-test)
331.17Average Return
4
Ant (test)
-2.54Best Test Reward
3
Ant non-stationary MuJoCo
18,095Final Training Reward
3
Ant Disabled
0.89Normalized Performance
3
Ant search phase
3.18D2C Score
3
Ant
960.57Average Decisions
3
Ant
0.82Action Repetition
3
Ant
60Normalized AUC
3